robot2
This page contains circuit diagrams and code for a walking robot created using Mark Tilden's nervous network, along with additional modifications like the reverser and brain circuits. The construction is unconventional, with chips having pins bent flat and components suspended within a space of approximately 1.2" by 1.5" by 2". The robot features a gearmotor at either end, a cordless phone battery mounted underneath, and feelers made from guitar strings. The overall length is about 4", excluding the feelers, and it has a leg span of about 3" in spacing and width. The robot can climb approximately 0.5" but does not perform well on carpet. The document includes various two-motor walker circuits developed by the creator. The first microcore prototype, which did not include motors, utilized a 74HC245 chip for motor driving but was never completed due to motor current limitations. The final design employed pairs of transistors to drive the motors, starting from a basic micro-core circuit modified with transistors. The need for reverse motion led to the exploration of using relays to swap motor connections; however, the difficulty in sourcing 3-volt relays and their high power consumption prompted the use of an analog switch chip instead. The initial choice of a 4016 chip was found unsuitable due to high on-resistance, leading to the use of two 4016 chips configured as a two-wire crossing circuit. This design proved effective, allowing the walker to retreat from obstacles. Subsequent discoveries indicated that stacked 74HC245 chips (or 74HCT245) were superior to transistor pairs for motor driving due to lower drive requirements and an enable input that prevents leg twisting during initialization. While transistor designs provide more feedback, they introduce a voltage loss that must be managed. True transistor H-bridges are avoided due to concerns about overheating. The documentation also includes a version of the Nervous Network, which is patented and prohibits commercial use without appropriate arrangements with the patent holder.
The walking robot circuit utilizes a microcontroller (PIC16C56) as the brain, which interfaces with two gearmotors to provide locomotion. The design incorporates a reverser circuit constructed from two CD4016 analog switches, allowing for directional control of the motors. The microcontroller manages the logic for forward and reverse motion, utilizing a series of neurons made from a 74HCT14 chip to process signals and control the motors effectively.
The circuit layout is compact, with components arranged to optimize space and performance. The motors are driven by the 74HC245 chips, which are capable of handling the required current while minimizing power loss. The enable input on these chips ensures that the motors remain in a neutral position during initialization, preventing mechanical strain.
The feelers, constructed from guitar strings, serve as sensory inputs, allowing the robot to detect obstacles and adjust its movement accordingly. The robot's ability to climb small obstacles is facilitated by the torque provided by the gearmotors, although its performance on carpet is limited due to reduced traction.
The overall design emphasizes modularity and adaptability, allowing for future enhancements and modifications. The documentation serves as a technical reference for those interested in replicating or building upon the existing design, while also highlighting the unique aspects of the nervous network concept as originally conceived by Mark Tilden.This page contains circuit diagrams and code for a walking robot I made using Mark Tilden`s nervous network along with my own tinkerings like the reverser and brain circuits. Construction is a bit odd, chips with pins bent flat with wirebits and parts in-the-air within about a 1.
2" by 1. 5" by 2" space with a gearmotor at either end, a cordless pho ne battery slung underneath and feelers made from guitar strings. The whole thing is about 4" long, not including feelers, with a leg-span about 3" spacing and width. It can climb about 1/2" and doesn`t do well on carpet. A little evolution. WALKER. TXT contains various two-motor walker circuits I devised. The first microcore I made (without motors) used a `245 chip for a motor driver, this walker never got made because my motors draw too much for it to power. The design that actually got made used transistor pairs to drive the motors. The circuit started out as a more-or-less stock micro-core circuit except for the transistors but I quickly realized the importance of being able to back up.
The obvious solution would have been a relay to swap the wires on one of the motors, however 3 volt relays are kind of hard to find and relays in general are power hogs. I tried using an analog switch chip to swap the driver signals, but the 4016 chip has too high of an on-resistance to drive the drivers (bet a 4066 would have worked and the design would be very different right now).
Then I thought, just make the pulse go backwards. I ended up using two 4016 chips each wired as a two-wire crossing circuit. This walker was pretty good and would back away from obstacles in its ways. [ Since this walker was made, I`ve discovered that stacked 74HC245 chips (or 74HCT245 as I used) work much better for motor drivers than complementary transistor pair drivers. The lower drive requirements of the chips make the reverser circuit simpler, and the `245 has an enable input that keeps the legs from twisting out of shape when initialized.
The transistors do give the microcore more feedback, if the 1. 3 volts loss isn`t a problem, but be sure the transistors can handle the motor current. True transistor H-bridges don`t have the loss problem but the threat of "smoke" has kept me from using them. See Another Nervous Net Walker for a circuit using the `245 chips. The design here is a good design if you desire lots of "feedback" and don`t mind the loss, but if building for the first time, don`t try to make this but use the much simpler new circuit.
This isn`t here for people to make anyway, but to document this particular robot - if you make it as-is you`ll be copying my less-than-efficient design decisions which may or may not matter. ] NOTE: The diagrams include a version of the Nervous Network, invented and internationally patented by Mark Tilden.
Commercial usage is prohibited unless arrangements are made with the patent holder. _ Rear Motor _ _: PIC brain : _ : Reverser : Drivers :_:. - Front Motor :-| | :_:_:_. -` <-`_ (angled) : | | : Microcore / Nu neurons : :Photo : : |_| :_: _ :_: :. |. |. :. `. . S-O Side : |_| :. -. : \_. -` `. Feelers : | : / : \ : Front (2) : <> : | Cordless Phone | `. <> : S Feeler : | : | (NiCad) Battery | `. \ : `. ` / `. ` | `-` \ \_| | \ | | | | |_| |_| Feel L Front Feel R - +| /- `-/-. photo / Nu- ->- = diodes : L R : : M = motors : : : *->-: 1-4 = Nv neurons (C input, R to gnd) *-:-:-:->-: Nu+ = pos. neuron (C to gnd, R to +). :. :. :. :. `-<-. Nu- = neg. neuron (R to gnd, C to +). -: `brain` :-<-: : :. : : Neurons made from 74HCT14 chip : :: + M - :: : Reverser made from 2 CD4016 chips : :`-2 4-`: : Brain made from PIC16C56 and 24C65 : : | /| : : Nu+ : | | :.
:. : : |/ | : : Reverser : Forward: 1->2->3->4->1 `->-`-1 3-` :. : Reverse: 4->3->2->1->4 + M - + | Npn >|-. Rev. Clr<-. . -|>|-*- +. -< Reset |e. -. | | /| |. -. | | Reset. -*-|470|-*-. *-O< |-*-|1. 5M|-` | 🔗 External reference
The walking robot circuit utilizes a microcontroller (PIC16C56) as the brain, which interfaces with two gearmotors to provide locomotion. The design incorporates a reverser circuit constructed from two CD4016 analog switches, allowing for directional control of the motors. The microcontroller manages the logic for forward and reverse motion, utilizing a series of neurons made from a 74HCT14 chip to process signals and control the motors effectively.
The circuit layout is compact, with components arranged to optimize space and performance. The motors are driven by the 74HC245 chips, which are capable of handling the required current while minimizing power loss. The enable input on these chips ensures that the motors remain in a neutral position during initialization, preventing mechanical strain.
The feelers, constructed from guitar strings, serve as sensory inputs, allowing the robot to detect obstacles and adjust its movement accordingly. The robot's ability to climb small obstacles is facilitated by the torque provided by the gearmotors, although its performance on carpet is limited due to reduced traction.
The overall design emphasizes modularity and adaptability, allowing for future enhancements and modifications. The documentation serves as a technical reference for those interested in replicating or building upon the existing design, while also highlighting the unique aspects of the nervous network concept as originally conceived by Mark Tilden.This page contains circuit diagrams and code for a walking robot I made using Mark Tilden`s nervous network along with my own tinkerings like the reverser and brain circuits. Construction is a bit odd, chips with pins bent flat with wirebits and parts in-the-air within about a 1.
2" by 1. 5" by 2" space with a gearmotor at either end, a cordless pho ne battery slung underneath and feelers made from guitar strings. The whole thing is about 4" long, not including feelers, with a leg-span about 3" spacing and width. It can climb about 1/2" and doesn`t do well on carpet. A little evolution. WALKER. TXT contains various two-motor walker circuits I devised. The first microcore I made (without motors) used a `245 chip for a motor driver, this walker never got made because my motors draw too much for it to power. The design that actually got made used transistor pairs to drive the motors. The circuit started out as a more-or-less stock micro-core circuit except for the transistors but I quickly realized the importance of being able to back up.
The obvious solution would have been a relay to swap the wires on one of the motors, however 3 volt relays are kind of hard to find and relays in general are power hogs. I tried using an analog switch chip to swap the driver signals, but the 4016 chip has too high of an on-resistance to drive the drivers (bet a 4066 would have worked and the design would be very different right now).
Then I thought, just make the pulse go backwards. I ended up using two 4016 chips each wired as a two-wire crossing circuit. This walker was pretty good and would back away from obstacles in its ways. [ Since this walker was made, I`ve discovered that stacked 74HC245 chips (or 74HCT245 as I used) work much better for motor drivers than complementary transistor pair drivers. The lower drive requirements of the chips make the reverser circuit simpler, and the `245 has an enable input that keeps the legs from twisting out of shape when initialized.
The transistors do give the microcore more feedback, if the 1. 3 volts loss isn`t a problem, but be sure the transistors can handle the motor current. True transistor H-bridges don`t have the loss problem but the threat of "smoke" has kept me from using them. See Another Nervous Net Walker for a circuit using the `245 chips. The design here is a good design if you desire lots of "feedback" and don`t mind the loss, but if building for the first time, don`t try to make this but use the much simpler new circuit.
This isn`t here for people to make anyway, but to document this particular robot - if you make it as-is you`ll be copying my less-than-efficient design decisions which may or may not matter. ] NOTE: The diagrams include a version of the Nervous Network, invented and internationally patented by Mark Tilden.
Commercial usage is prohibited unless arrangements are made with the patent holder
:. : : |/ | : : Reverser : Forward: 1->2->3->4->1 `->-`-1 3-` :. : Reverse: 4->3->2->1->4 + M - + | Npn >|-. Rev. Clr<-. . -|>|-*- +. -< Reset |e. -. | | /| |. -. | | Reset. -*-|470|-*-. *-O< |-*-|1. 5M|-` | 🔗 External reference